YMatrix is a distributed database product developed from the open-source PostgreSQL/Greenplum ecosystem, featuring the following key characteristics:
In addition to the commercial edition, YMatrix also offers a free community edition. Your experience and feedback are welcome.
A hyper-converged database integrates transactional (OLTP), analytical (OLAP), time-series, and data lake capabilities into a single database system.
YMatrix’s hyper-convergence philosophy eliminates data processing fragmentation by unifying compute, storage, and network resources within one system. Based on original database types, versions, cluster topology, and business characteristics, YMatrix delivers tailored combinations of storage and execution engines atop a shared database foundation. This enables specialized micro-kernels optimized for write, storage, and query performance across diverse business scenarios.
YMatrix emphasizes full-scenario functionality and performance, including ingestion, querying, analytics, and machine learning. By integrating multiple capabilities into a single database, it addresses complex use cases and achieves multi-model support, scalability, and cost efficiency.
· Analytical Capabilities
· Transactional Capabilities
· Time-Series Capabilities
YMatrix uses SQL as the unified interface for all data services at the application layer.
YMatrix provides strong extensibility.
On one hand, YMatrix continues to expand into new business scenarios such as connected vehicles, smart manufacturing, finance, and vector data processing. On the other hand, through features like machine learning and federated data access, heterogeneous and external data sources can run efficiently within YMatrix via database extensions.
By simplifying infrastructure architecture, YMatrix significantly reduces technology stack complexity, improves performance across diverse scenarios, minimizes risks from multi-system coexistence and integration, and helps enterprises build robust data governance frameworks—unlocking the full digital potential of the data era.
YMatrix leverages several key self-developed technologies to realize its hyper-convergence vision.
Designed for OLAP, OLTP, and time-series workloads, MARS3 offers two modes: columnar and hybrid row-columnar storage, allowing users to choose based on workload needs. The hybrid mode ensures both high ingestion performance and efficient storage (including compression and health diagnostics). Both modes implement MVCC. For partitioned tables, MARS3 supports automatic partition management and automatic storage tiering.
The vectorized execution engine is designed specifically for column-oriented storage engines such as MARS3, MARS2, and AOCO. It delivers one to two orders of magnitude better performance than traditional row-based execution engines for common queries.
ALOHA (Advanced Least Operation High Availability) is a cluster state management service introduced in YMatrix 5.X. Running independently from the main cluster, ALOHA can be configured with dedicated disks and monitoring. It ensures low-latency node status detection and management even under harsh conditions, completing automatic failover within 3 seconds.
Powerful analytical computing capabilities
Traditional data warehouse workflows rely on the Hadoop ecosystem: storing historical data in Hadoop and using Spark for report computation—resulting in complex pipelines.
YMatrix resolves this complexity through hyper-convergence while enhancing analytical performance. By supporting structured and unstructured data types and federated data access, YMatrix handles BI and reporting tasks in classic OLAP scenarios across finance, telecom, government, energy, and manufacturing. Advanced query optimizations such as vectorization, Runtime Filter, sliding windows, and continuous aggregation deliver superior analytical performance.
Balancing high-speed ingestion, low-cost storage, and real-time querying
Time-series data demands high performance in ingestion, storage, and querying due to its real-time nature.
YMatrix is optimized for time-centric workloads. Thanks to physical time ordering in the MARS storage engine, asynchronous and batched uploads, and MatrixGate’s high-concurrency, high-throughput ingestion, YMatrix exceeds expectations in real-time data loading, querying, and transaction guarantees.
YMatrix supports graphical, non-disruptive expansion—enabling simple, second-level scaling without service interruption, ensuring business continuity, minimizing downtime costs, and reducing operational risk.
Leveraging hyper-convergence to unify data pipelines
Data silos are common in traditional enterprises. Isolated data cannot be shared or utilized effectively, hindering management, operations, and growth—and blocking digital transformation.
YMatrix’s hyper-converged architecture has been successfully deployed in real-world production environments such as factory data platforms, enterprise group data warehouses, intelligent connected vehicles, and IoT device operations. It significantly lowers technical barriers in selection, procurement, deployment, and maintenance. For example, in smart manufacturing, a single YMatrix instance can collect, store, compute, model, query, and analyze data from ERP, MES, and equipment systems—all within one database.
